Ah yes, but... [LWN.net]

Ah yes, but...

Posted Oct 20, 2006 20:48 UTC (Fri) by jd (guest, #26381)
Parent article: Thousands of TeX fonts at your fingertips (Linux.com)

A lot of fonts on CTAN are not of the same quality as the default fonts. As such, their right to be included as a "real" font is dubious. (I would say the same about any font collection.)
The range of historic and foreign language fonts is definitely limited at this time.
Although the rendering is vastly better than that of any of the "popular" font systems, the metafont system lacks some of the subtleties present in modern font formats. This isn't a massive penalty, as nobody uses them anyway, but a limitation is a limitation no matter how you cut it.
Although metafonts are generated algorithmically, code generators or IDEs that produce metafont algorithms are somewhere between scarce and non-existant. Hand-coding is, of course, always possible if there is sufficient data to produce a mathematically-correct description.
Although metafonts ARE vastly superior (when produced well), they are only really used by TeX. Which is odd, as they would allow web pages to embed fonts in a platform- and resolution-independent format that meets the HTML ideal of describing a page textually.

For these reasons, statements such as "thousands of free fonts" mean as little for metafonts as for truetype. This should not be the case - the fonts should be vastly more versatile, vastly more amenable to analysis, design and development, and vastly more widespread than any other font system.

Yes, I am frustrated by the agonizingly slow pace of development on metafonts, TeX and LaTeX (which will reach version 3 any century now), and annoyed by the almost total isolation the branch has moved into. This is an area that has the attention of a decent chunk of the scientific and academic press (none of whom are rich, but none of whom are poor, either) and an area simple enough for Universities to assign development work for across a whole host of departments.

By sheer weight of potential numbers and potential cash, this should be the single-fastest developing region of Open Source. By sheer availability, it should also be the most international of all document development environments and the most widely used by archaeologists and anthropologists who need to be able to describe written, spoken or gestured language in rare or unknown forms right there and then.

I couldn't tell you what the underlying problem is, only that it is self-evidently great enough that one of the best font systems ever devised is totally unused and unknown outside of TeX.

(Log in to post comments)

Ah yes, but...

Posted Oct 20, 2006 21:24 UTC (Fri) by hanwen (subscriber, #4329) [Link]

FWIW, LilyPond also uses MetaFont and is (by now) fully decoupled from TeX.

It's a debatable whether MF is a good environment for designing fonts, though. Font design is largely visual, while parametric description of curves requires some mathematical cluefulness. The combination seems very rare in font designers. I think the TeX CM series (largely written by Knuth himself) is the only real font where each and every parameter is tunable.

Ah yes, but...

Posted Oct 20, 2006 23:54 UTC (Fri) by jwb (subscriber, #15467) [Link]

Most of what you say is true, but MF is, as far as I can tell, incapable of dealing with a low-resolution device such as your monitor. Other font systems are specifically designed to cope with 75dpi. MF looks wonderful in 2400lpi offset lithography, or 600dpi laser scanning. There's a major difference.

Ah yes, but...

Posted Oct 23, 2006 10:47 UTC (Mon) by anselm (subscriber, #2796) [Link]

I think you're confusing Metafont with Computer Modern (Knuth's exemplary
font family, implemented in Metafont). Metafont does contain features
that allow a font designer to cater for low-resolution devices (it is all
in the documentation). However, a font like Computer Modern Roman, by
design, only looks good at very high resolutions. Even 600dpi does not
remotely do it justice -- compare a CMR laser printout to, say, the
TeXbook.

Anselm

Ah yes, but...

Posted Oct 28, 2006 0:54 UTC (Sat) by jschrod (subscriber, #1646) [Link]

Please note that most fonts in TeX distributions are *not* MF fonts, but are Type1 fonts. Even on CTAN, most MF fonts are symbol fonts, we have very few type families that are made in MF.

Also, the TeX community goes away from MF, and targets Type1 (with the GUST work) or Opentype fonts (the latter with XeTeX and new work on Omega that will hopefully get merged with pdfTeX).

And, please don't bring up the LaTeX3 straw man. It's a system to experiment with typesetting and a new way of style programming. Current development for authors happens in LaTeX packages and not in the core system -- and there a lot happens, as one can see from the CTAN announcement list. Of course, there is also ConTeXt, for those who are more interested in powerful typesetting (see http://www.pragma-ade.com/overview.htm).

Joachim
[from the CTAN team; co-author of the LaTeX Companion, 2nd ed.]